serverless in digital landscape an infosys point of view · source serverless platform based on...

VIEW POINT

SERVERLESS IN DIGITAL LANDSCAPE AN INFOSYS POINT OF VIEWBy Archa Shukla

Introduction

Executive Summary

are emerging in the FAAS space that

provides Function portability and more

flexibility in terms of development .Few

of the promising ones are Kubeless,

Fission, KNative , OpenFAAS. These can

be deployed on Kubernetes, a popular

container orchestration platform.

Enterprise who choose to take the

Serverless path can pick today from these

two options:Vendor provided Serverless

Cloud platform or Vendorless Open

Source Serverless Platform based on

their specific need.

role in its Digitalization journey. Serverless

should be integrated as a core part in any

Enterprise Digital Strategy . We suggest

the use of Serverless in building Customer

Experience Platform and extend the

discussion further to embed IoT data in

the landscape for a complete connected

user experience. We present the reference

architecture and some of the complex use

cases where Serverless architecture can

benefit the Enterprise and help it address

the need to quickly scale the systems for

rapidly increasing load by ever increasing

Consumer base of digital services.

Architecture of present day leading edge

applications need different ways of data

interaction and requires more adaptive and

dynamic capabilities. Hence, the need of

the hour is to move these implementations

from a monolithic architecture to a Micro

service based architecture that is event

driven and more data centric. Functions

as a Service approach take micro services

approach to next level by providing smaller

lightweight, independently scalable

and fully managed functions as services

(Functions as a Service) that are Event

Driven. These functions provide basic

Digital landscape is ever growing and

evolving in order to meet the changing

market demands. Internet of Things,

Blockchain, Augmented Reality, Virtual

Reality, Robotic Process Automation and

other advanced technologies are shaping

the future of enterprises and driving the

need for a Digital transformation work.

The real time information provided by

IoT is being utilized in many Mobile, Web

applications and is deployed in almost

all the verticals namely Energy, Telecom,

Agriculture, Manufacturing, Health and

Retail. More and More customers want

building blocks for providing a micro

service based architecture.

Pubic Cloud vendors have created an

Infrastructure to provide Functions as a

Service with Fully managed infrastructure

and named them as Serverless Functions.

All the major cloud vendors today are

providing Serverless functions for use.

Some of the widely used Serverless

functions we know today are AWS Lambda

Functions, Microsoft Azure Functions,

Google Cloud Functions and IBM functions.

On the other hand, Open source platforms

to be digitally connected using the latest

technologies and this is creating a need for

dynamically scaling the Enterprises Digital

Landscape. Enterprises today with digital

offerings see an ever-increasing demand

for agile development environment, high

scalability and availability to meet the

ever-increasing Customer demands for a

digitally connected world.

In this context, we present our point of

view that addresses this current need and

explains how Serverless architecture can

help in modernizing the current digital

landscape of an Enterprise and play a vital

Customer Experience Layer (CMS,

Angular JS , Xamarin, iOS, Android ,

Cordova, React , ReactNative)

Serverless Business Logic Layer

Serverless -Vendor Specific and Open Source Options

Web and Mobile

Customer

Experience Layer

AWS Lambda OpenFAASAzure

Functions KNativeGoogle Functions FissionIBM

Functions Kubeless

Vendor specific Serverless functions

on Public Cloud

Open Source Serverless functions on Kubernetes , Docker and Container

Registry

Web and Mobile

Customer

Experience Layer

External Document © 2020 Infosys Limited External Document © 2020 Infosys Limited

Serverless: The Future of Digital Landscape

adversity it is unequivocal that the fate

of the Digital space is Serverless and an

ever-increasing number of organizations

ought to acknowledge this change and

consolidate Serverless as an essential piece

of their Digital Strategy.

In a ‘Digital First’ world, the Customers,

Partners and Internal users of an Enterprise

need all services via digital touchpoints to

self-serve themselves, anytime, anywhere.

The demand for faster development and

quick deployment of features is rapidly

growing. The applications need to be

scalable and agile. By combining Serverless

functions and other cloud services,

Enterprises can build the most powerful

web and mobile applications that can

scale up and down based on the demand.

Key areas that we may benefit from

Serverless are in the space of Customer

journeys involving marketing content and

e-commerce touchpoints.

Netflix (Refer Credits) is one of the biggest

Industry leaders and trendsetters in using

Innovative and new technologies. Netflix

has moved its popular Video Content

Streaming site from On Premise Data

Centres to a completely Cloud Native

approach. It is intensely utilizing Serverless

functions to accomplish Media Content

Sorting, deploying, publishing, approval

workflows, monitoring using Lambda and

other connected AWS services. Without

Serverless, it was difficult to achieve such

a highly scalable and robust platform to

server millions of Subscribers.

ShutOut (Refer Credits) , which is a Stratup

and offers a SAAS platform for improving

Customer Engagements, Communication

and Loyalty Management platform have

recently transformed their platform using

Serverless architecture. This provided them

huge benefits to deliver innovative ideas

faster than ever and at scale with reduced

costs.

Some of the other major players in the

market who have embraced Serverless and

reaped great benefits are Coca Cola, BBC

and AIRBNB.

Coca Cola (Refer Credits) reduced its

Operational Costs by going Serverless.

BBC (Refer Credits) has created a Serverless

Mobile platform based on AWS Lambda

and API gateway.

AirBNB (Refer Credits) has designed its

own Alert Streaming framework using

Serverless architecture.

With all these examples of overcoming


Recommended Approach to Modernize an Application using Serverless

Recommended Migration Approach

code is then migrated to the Serverless

architecture. In this move from Monolithic

to completely Serverless FAAS, there will

be a phase where the application will work

in a hybrid mode that despite everything

will have numerous advantages over

completely monolithic system.

An enterprise on an existing platform

that is on a monolithic architecture and

confronting issues regarding Granular

scaling of the application, Higher

Operational administrative expense and

lower flexibility. The recommendation

is to start thinking in the direction of

modernizing its landscape using Serverless

and event driven functions.

The critical approach is to isolate modules

that encompasses the core of the business,

yet can be modularized and hosted as a

set of functions that can scale by itself.

Rather than adopting a big bang migration

strategy, the recommendation is to go for

a simple stepwise, iterative methodology

in which bottlenecks in the present

functionalities are identified .Selected

Monolithic Hybrid (Monolithic + Serverless) Serverless FAAS

Assessment and Analysis

of As Is State

• Analyze Existing

bottlenecks

• List candidate

functionalities for

migration

• Prioritize Functionalities

to be migrated

Optimization

• Monitor these functions

for the benifits gained

• Uitilize tools and

mechanisms for

Observability of these

functions

Analysis of To Be State

• Identify TO BE Serverless

platform

• Open Source or Vendor

Specific

• Identify Serverless

Observability tools

Migration and Testing

• Implement , test and

deploy final functions

• Start migration of other

functions based on

priority

Automation

• Automate Deployments

using Infrastructure as

Code

• Automate Unit testing

and End to End event

driven service testing

Proof of Concept

• Select a Functionality for

migration

• Create Proof of Concept

by migrating the code

• Test the concept


Serverless Architecture for Core Digital Platforms

• RDS database

• Cognito for authentication and

authorization.

• S3 for static content host

Customer Experience Platform

The Customer Experience layers for the

digital applications is moving from heavy

weight applications / portals we have seen

in the past to lean user interfaces or lean

portals. These are built using frameworks

like Angular, React, HTML5, JS, React

Native etc. The present need is to keep

the logic at user interface very light so

that the developers could focus on only

the interface design and create best in

class user experience without bothering

much about managing and deploying

the business logic layer. All the business

logic need to be moved to a separate layer

that could be reused across channels,

managed and scaled independently of

the presentation layer. Lightweight User

Interface can consume backend functions

and business logic using APIs exposed via

API gateways. Most of the services available

in the cloud now are HTTP enabled and

the Customer experience layer can interact

with these Cloud services via JWT tokens/

IAM roles or other similar mechanisms. This

makes even easier to develop an entire

application using a light weight Customer

experience layer and a combination of

Serverless functions in the backend and

integration with other Cloud services for

Authentication, Payment, Virus Scan, File

Upload /Download (BAAS –Backend as a

Service)

Bustle.com(Refer Credits), a news

entertainment and fashion website has

taken a major initiative to migrate all

their Customer experience journeys on

Mobile and Web from On Premise to Cloud

based Serverless architecture. The new

architecture is based on AWS lambda and

other connected services.The company is

gaining tangible benefits from the same.

A simplistic view of such a Serverless

layer is shown below .This could be

implemented using Vendors provided

Serverless platforms or Vendorless

Serverless platforms.

A typical implementation can use the

following services from AWS Cloud

platform –

• Lambda and Step functions

Serverless Customer Experience Platform

Serverless functions Transactional Database CRUD operations, Data transformation functions,

Notifications

AA

A

Relational Database

API Gateway

Customer Experience Layer


Content Management Systems

systems and after reaping the said benefits

and proving their Business Case, they can

slowly start replacing other systems as well.

Altassian (Refer Credits) is a software tool

provide company with Australian roots and

Global reach.

Atlassian has recently leveraged

Contentful to manage content for a

growing list of external support services:

help and technical documentation, FAQs,

help articles, product documentation, in-

product support, and resources for people

considering buying Atlassian products.

Contentful is a SAAS solution that offers

Headless Serverless platform. It handles

spikes in traffic, manages scalability,

flexibility to add third party integrations

and expansion to Global markets.

2. Small and Medium businesses and

Stratups

Small and Medium businesses, who want

to start small initially, should consider

build a basic Content Management System

from scratch using public or open source

Serverless options. They can also consider

Open source CMS systems available in the

markets that run on Serverless like Aerosol

CMS and Webiny.

Such businesses can implement Static

content hosting using object stores like S3

buckets and deliver them using Content

Delivery Networks (CDN).More complex

features like content authoring, inline

editing, content workflows, themes /

templates , User segmentations can be

implemented incrementally as and when

the business grows using a well thought

design and Serverless backend.

3. Hybrid Approach

Companies currently using a Headless CMS

approach but facing some challenges in

terms of scalability can start modernizing

some of their functions using Serverless

approach.

The legacy Web CMS used by some

Enterprise today are based on a Monolithic

architecture and have their Presentation,

Content Management and Content

Delivery systems tightly coupled. These

pose various challenges as stated below:

1. Channel- Centric

Earlier CMSs were built for websites and

remain very channel centric. Because of

this, Enterprises had to use different CMSs

for multiple Channels.

2. No on-demand scaling

These CMSs have heavyweight ecosystems.

They do not scale very well when needed.

3. Offer No or minimum agility

These systems do not offer agility in term

of delivering content faster to the market

and have rigid workflows.

4. Extensibility and Integration

Integration with other tools in the

enterprise systems like PIM, Portals,

Personalization etc. become difficult.

Because of these challenges, product

vendors and companies are exploring

various ways to implement lightweight,

Micro functions based CMS systems that

also scale on demand. Headless (API first),

Serverless approach fits best here.

Many Product vendors are embracing this

need and have some good offerings in this

area like Contentful, Webiny and Cosmic

JS. These provide Headless Serverless CMS

systems.

Headless systems provide API-first

approach and provides a decoupled

solution for Content Presentation and

Management. Serverless take this to the

next level and provide on demand scaling

and lower Operational admiration costs.

Some of the CMS patterns to address this

need are:

1. Large Legacy Enterprises

These Enterprise want to move away

from monolithic Web CMS system can try

modernizing their stack using

Headless Serverless CMS systems .They

can try to pilot a green field project using

the available Headless Serverless CMS

Headless - Serverless CMS Approach

Authoring Environment Live Environment

Business Authoring Site Customer Experience Layer

Serverless Content FunctionsContent Read , write , update , publish from stage to live , workflows

CDN

AA

A

AA

A

Content Stage (version enabled

Object store)

Content Live (version

enabled Object store)

Content Metadata and

persistent data (Database)

Content Metadata and

persistent data (Database)

API Gateway


E-Commerce Systems

some gaps and hence can embrace a

Serverless extensions against an extension

on existing Vendorless platform itself.

An example use case in the Hybrid

approach is building a Global Inventory

view for Retail sector in near real time .This

will help in implementing Omnichannel

fulfilment capabilities.

The headless Serverless MVP can be

easily achieved using a combination of

the following:

Lambda Step functions, No SQL DB like

Dynamo DB and S3 object store. AWS

Cognito can provide Identity and Access

Management.

E-Commerce sector is facing major

business challenges today. The companies

need to meet high customer satisfaction,

lower wait times on site, and reduce the

wastage of resources during downtimes,

scale during peak hours and all this with

reduced administration and operational

cost. Few of the challenges faced by an

e-commerce business are:

Need On demand scaling

These systems have fluctuating demands

throughout the year and have a great

increase in the demand during special

occasions like Black Friday, New Year, and

Christmas.

In a traditional monolithic approach,

the Enterprise has to plan for the peak

time capacity and use the required no.

of instances. Even if they support auto

scaling, individual and granular scaling of

features is not supported.

Lower Operational Cost

Enterprises usually purchase the systems

today from popular product vendors and

have to pay overwhelming license cost .On

top of the licenses cost the Administrative

Operational cost and Administrative efforts

of such monolithic traditional systems is

another cause for concern.

Reduce Time to Market

Any new ideas in a monolithic application

takes time for development and testing

and hence, the overall time to market

increases.

Some e-commerce patterns to address these bottlenecks:

1. Large Legacy Enterprises :

Large enterprises leveraging vendor

based COTS products need to reduce

Cost and increase scalability can consider

modernizing their stack by going

Serverless as explained earlier in this

document.

MatHem (Ref. Credits), Sweden’s largest

online grocery store has recently

redesigned their site using AWS Serverless

functions. This helped them in achieving a

highly scalable on demand web site .Also,

lowered the time to market for new ideas

and reduced overall operational cost.

2. Small and Medium Businesses and

Startups:

Smaller businesses and Startups, who

cannot invest huge amounts in purchasing

a COTS e-commerce product, can build up

their own custom online shop platform

easily using the Serverless approach.

3. Hybrid Approach

Vendorless Headless approach may have

Headless Serverless e-Commerce

Serverless functions Products , Cart , Inventory , Payments , Order, Notifications, Customer

support ,Reporting

AA

A

No SQL DB (Product hierarchy , Cart etc.) Object Store Product Images, manuals , videos

API Gateway

Customer Experience Layer


Reference Conceptual Architecture

- Core functions comprising of Content

Management , e-commerce and

Customer Experience Data base

functions ( Lambda functions with ECS

tasks)

- Enabler functions comprising of Data

transformation, processing functions

and Notifications. (AWS Glue, Lambda,

Kinesis, ECS etc.)

A consolidated reference architecture that

shows the use of Serverless Layer in an

Enterprise Digital Landscape is illustrated

below.

Serverless functions play a key role

and help us in implementing the new

Serverless function layer that implements

business logic for applications. These

Serverless functions act as micro functions

and expose backend functionalities as

micro services. Each functional unit can

be developed, deployed and scaled

separately.

The layer can be divided into the following.

- Function Invocation Layer (Invoked from

API gateway or triggered based on an

event from any other Cloud service like

Lambda, Kinesis, SQS, SNS S3 etc.)

- Function Workflows and Orchestration (

Lambda step functions )

Frontend Channels (Desktop , Mobile , Tablets etc.)

Enterprise Systems Integration Middleware

Serverless Functions Layer

Core Functions

Enabler Functions

Operational and Referential Data

Store for Customer journeys

Data Processing /Transformation• Real Time streaming data

transformation• Real time ETL operations• Real time file transformation• IOT data processing

Real Time Notifications• Mass emailing• Mass SMS• Send Alerts to DB or

Mobile notification services

E-commerce Functions• Shopping cart• Product and Pricing• Inventory• Payment• Order management• Fulfilment

Content Processing /Delivery• Authoring Workflows • Live content delivery• Content Authoring• Content CRUD operations• Content Publishing

Transactional and reference data operations• Create• Update• Delete• Read

Data Monitoring and Analytics functions• Raw data conversion

, filtering and transformation

Synchronous /Asynchronous Triggers

Step Functions for Orchestration and Workflows

File and other static content

store

Notifications from Enterprise systems

Secu

rity

(AA

A)

Dat

a A

naly

tics

Real Time Streaming data Data LakeIOT Data

API Gateway

Lightweight Customer Experience layer(Content Management Systems , Enterprise Portals , Mobile Applications, e-commerce Sites)

Enterprise systemsCustomer Data, Product Information Management, Commerce, ERP , Master Data

Serverless Functions Layer ( Function as a Service, Event Driven, Auto Scalable)


Service Mapping for Serverless Layer and Supporting Functions

Note: The services are only indicative .Every Cloud offering has numerous services for similar functionality and should be evaluated based on

the Enterprise need

Public Cloud Vendors Open Source

Layer in Reference Architecture

Services AWS Microsoft Azure Google IBMKubeless / Open FAAS

Security Authentication Cognito Azure AD Google Firebase Identity ServiceFlexibility to use any Open Source , Public cloud services

API Gateway API Gateway AWS API gateway API ManagementGoogle Cloud Endpoints

API ConnectFlexibility to use any Open Source , Public cloud services

Serverless Function Layer

Serverless Functions Layer

AWS Lambda Functions Cloud Functions Open Whisk

Kubeless /Open FAAS functions deployed on Docker Container

Data Layer Object Storage S3 Blob storage Cloud Storage Object StorageFlexibility to use any Open Source , Public cloud services

Data LayerRelational data storage

RDS Azure database Cloud SQL Data ServicesFlexibility to use any Open Source , Public cloud services

Data LayerData Layer: Non Relational Data storage

Dynamo DB Cosmos DB Cloud Data Store CloudantFlexibility to use any Open Source , Public cloud services

Data Layer IOT data AWS IOT Azure IOT Google IOT IBM Watson IOTFlexibility to use any Open Source , Public cloud services

Data Layer Notifications AWS SNSAzure Notifications Hub

Cloud Messaging IBM MQFlexibility to use any Open Source , Public cloud services

Big data and Analytics

Data AnalyticsRedshift Kinesis Elastisearch

Synapse Analytics

Stream Analytics

Google Analytics

Cloud Data FlowSmart Analytics

Flexibility to use any Open Source , Public cloud services, Kafka event streaming

Monitoring MonitoringCloud Trail, Cloud Watch and Third party

Operational Insights

Cloud ConsoleMonitoring and Reporting

Flexibility to use any Open Source or other tools


Business Benefits of Proposed Serverless Architecture

server software /hardware management is

taken care of by the Cloud Vendor hence

reducing Operational Administrative

efforts and cost.

• Flexibility to choose development

language

Serverless functions support multiple

languages and provide a choice of

language for implementing each use

case .Some of the supported language

provided by Lambda functions are Java,

Go, PowerShell, Node.js, C#, Python, and

Ruby code, and provides a Runtime API

which allows you to use any additional

programming languages to author your

functions.

• Connected Cloud Services and Event

Driven

Serverless functions are very well

connected to other Cloud services in the

ecosystem. This enables the architects

to tackle many complex scenarios that

may be encountered in a complex digital

landscape.

For example -Lambda functions can

be triggered synchronously and

asynchronously by multiple AWS services.

• Integrated Security

Serverless function providers have

implemented the security using IAM

policies at very granular level.

The following is controlled via Security

policies

- The access and trigger to these functions

- Access to other services from these

functions

• Operate in Multi Cloud Environment

Enterprises today want to utilize the best

services from different cloud vendors.

They would want to utilize AWS API with

Lambda and Google IOT.

In these scenarios, they have flexibility

to deploy a mix and match of lambda

functions in multi cloud environment.

The proposed architecture serves following benefits

• Flexible and Adaptive

Creating a Serverless functions layer

provides flexibility to independently add,

remove and update newer logic based

on the current business requirements.

Serverless functions provide capability

to be invoked synchronously or

asynchronously based on certain events.

These can be exposed via an API gateway

for consumption or can be invoked based

on certain events such as Data change

events in DB, File upload/modify events,

Message Creation/Deletion Event in

Message Brokers

• Agile and faster development

Functions are smaller and can be

developed faster, deployed independently

and can be added any time based on new

business needs.

• Separation of Concerns

Each layer depicted in the proposed

architecture addresses a specific need .The

Serverless Function layer focuses on data

processing and providing Business logic for

the presentation layer

• Reusable Business Logic

The business layer functions are reusable

and can be consumed by all the User

touchpoints /channels via APIs.

• Highly scalable and Available

Serverless architectures can build powerful

web applications that can auto scale up

and down based on the demand and run

in high available cloud environments.

Scalability, backups are taken care of

by the Cloud provider. Each function is

independently scalable. For example

Lambda functions can auto scale. Based

on the no. of requests received by each

lambda function it can increase the no. of

instances running or reduce the instances.

• Low Operational Administrative

efforts and Cost.

Serverless functions do not need server

software or hardware management by the

enterprise or developer. It provides a Pay

per Use model and does not incur any cost

when the function is not being used. The


Complex Usecase Implementation in Serverless World

The following key Customer Experience

journeys for the end customers can be

envisaged which can be achieved using a

Serverless architecture.

databases for further analysis and real

time display in dashboards. The telemetry

data example are historical machine track,

Machine Hours or Machine Fuel value.

Alerts and notifications can be generated

from this data to inform the user about

specific actions to be taken on the vehicle/

equipment. Also, auto actions can be taken

by ordering a new equipment or part,

which needs urgent replacement.

Consider the applicability of Serverless

architecture with an intricate real life

example.

A Manufacturing Enterprise that builds and

sells agricultural devices and equipment

Sample implementation of the above

journeys -

1. Use Case 1: Machine Telemetry data Stream processing (Building a Real time ‘Extract Transform Load’ pipeline)With the increase in connected devices

and the need to monitor these in Real time

in applications, we believe that IoT data

Connected Device Lambda FunctionKinesis Stream

Trigger Lambda

Raw IoT Data (CAN BUS data)

Web/Mobile app Database (Operational and

Referential data store)

Convert Raw data into required format and store

in DB

Predefined condition

Database

including connected Machines i.e.

tractors and implements. The company is

embarking on a Digitalization journey and

wants to provide Digital Touchpoints to its

consumers via a Self-Care Web and Mobile

Channel.

storage and processing is an important

feature where Serverless functions may

play a key role.

Machine / Connected device telemetry

data received via a CAN BUS and be piped

into Kinesis Streams and can then be

processed using lambda. This data can

be analysed, filtered, transformed and

stored in Data Lakes and Application

Customer Journey Serverless Architecture Applicability

Use Case 1: Machine Management Viewing machine fleet and Monitoring the machine telemetry data by processing real time IOT data

Real time IOT data processing and storing in Data store for display in the Customer Self-care site.

This data could be anything from current machine location, machine hours , fuel level etc.

Use Case 2 : Field and Task data Management and Processing Processing of agricultural data related to Fields and Tasks to be executed on fields.

View Access Field and Task data from Data Store

Converting and storing data received by Machines (recorded fields and completed tasks)into data format that can be stored in data store of the Customer self-care site and also send the updated data(updated fields and created tasks) from data store to Machines

CRUD operations on the data store that have to be exposed to the portal. The data includes Field data, task data and Machine telemetry or any other reference data.

Use Case 3 : Using data analytics to improve portal experience e-commerce journeys to buy Machines and Accessories

Use data analytics to improve customer experience

Use Case 4 : Receive Notifications Send Machine specific notifications or any other alerts relevant for the user.

Use Case 5 : Shop Online for Machine Parts An after sales online shop that offers purchase of tractor parts.

Real Time Telemetry Processing

Use Case Fig: 1


2. Use Case 2: Field and Task Data Management and Processing

a. Database Operations

Storage of Operational and Referential data is a key use case in any application. Serverless functions can help exposing this data in the form

of micro functions to provide all CRUD operations.

In our sample, business scenario Operational and Referential data store is used to store processed Field, Task and Machine Telemetry data.

2. Use Case 2: Field and Task Data Management and Processing

a. Real Time File transformation functions

Raw data obtained from agricultural data exchange platforms could be placed in S3 buckets. Every write, update in S3 auto invokes a Lambda

function.

These lambda functions can call external converters that help to convert the agricultural data files to specified format and vice versa. The

converted files can be stored in Processed files S3 buckets.

Lambda FunctionAPI Gateway Data CRUD Operations

Web/Mobile app Database (Operational and

Referential data store)

Web/Mobile app Database

Web Application / Mobile Application

Database

Database

Database Operations

Use Case Fig: 2

Trigger Lambda

Connected Device Lambda Function

Lambda Function

Trigger Lambda

Trigger Lambda

Raw Agricultural data in Files

Data transfer via Agri data

exchange platform

Convert and Store in

Database

Processed Data Files

Invoke External Utilities

File Converter Utility

Real Time IOT data and File Processing

Use Case Fig: 3

S3 Bucket S3 Bucket


Trigger Lambda

Trigger Lambda

Connected Device

Lambda Function AWS

RedshiftData Analytics

And Dashboards

Lambda Function

Data Transformation

Raw DataData Warehouse Business Intelligence

Tools

Social Media, Clickstream

Enterprise system Database in AWS

(Order and Customer Information)

Data and Analytics

Use Case Fig: 4

S3 Bucket

vehicles and improve their processes.

Serverless lambda functions help here as

they can be easily triggered by Streaming

data and S3 buckets .Lambda functions

can be used easily to process this data

, transform and load into data lakes.

Business Intelligence tools like AWS Quick

Sight , Tableau are used to analyse this data

and present the results to Business users in

Interactive Dashboards.

3. Use case 3 : Data AnalyticsData Analytics is a must have functionality

for any Enterprise today. Analytics help

the Enterprise improve Operational

efficiency, drive new revenue and gain

competitive advantages over business

competitors. Various forms of data is

analysed today by any enterprise to gain

important insights about their customer

and products. This includes streaming data

coming from Social Media, Telemetry data

from Connected Devices and Vehicles,

Website clickstream ,Customer Orders,

Product Usage data . This data needs to

be extracted from their source systems,

transformed, sorted and loaded in Data

Lakes. The Analysis of data and use of

predictive modeling tools help companies

to understand their customers , improve

user experience of sites, do targeted

promotions and run marketing campaigns ,

avoiding equipment failures , servicing the

Kinesis Stream


Mobile and Web Application Notifications

From SNS these messages can be pushed

to Mobile via platform specific notification

services like Apple Push Notifications and

Google Push Notifications and to web

application via inserting in application

Database. Another set of lambda functions

can be leveraged to provide some post

processing on the messages like applying

templates for eMail and SMS.

4. Use Case :Notifications to Web and Mobile UIPush notifications, e-mails and SMS are

effective tools to keep customers engaged

with the applications and inform them or

warn them about things that need their

attention.

Lambda functions could be used to

trigger such notifications. Example of

such a notification is the Expiry of Vehicle

warranty, Low Fuel level in Vehicle, Order

dispatched, Order Delivered , New updates,

Service Contact Expired etc.

Data coming from IoT data streams or

certain events in other enterprise systems

like Order Creation, Order dispatched, Task

Completion, Task Assigned can trigger

lambda functions .Lambda functions

can filter the required events and create

required Notifications messages in AWS

SNS service.

Trigger Lambda

Trigger Lambda

Connected DeviceLambda Function

Lambda Function

SNS Topics

Data Warehouse

Raw IoT Data

Analyse data and Generate Alerts/

Notification Messages

Use Case Fig: 5

Kinesis Stream

Mobile Native Apps

Mobile Platform Push Notification services

Enterprise Systems

Web/Mobile app Database Responsive Web Application

Database


5. Use Case : On line Shop for purchasing Machine PartsAn engaging e-Commerce site is necessary for any Manufacturing company. This can be easily implemented using a combination of Object

stores, No SQL , SQL databases and lambda functions.

A simple use case of ‘Fetching the Product hierarchy’ is provided below.

CDN

Lambda FunctionAPI Gateway Get Product Hierarchy

View Product Catalogue

No SQL Dynamo DB: database to Store Hierarchy

and Product Prices

Web Application / Mobile Application

Database

Online Shop

Use Case Fig: 2

Trigger Lambda

Raw Agricultural data in Files

Get Product Images

Product Image Storage


Critical Challenges faced in Serverless Approach

of around 15 minutes. This limitation can

be easily overcome by leveraging Step

functions to break down the processing

logic into multiple steps and combining

with ECS tasks. Lambda will work as an

invocation layer and the actual heavy

workload could be handled by the ECS

tasks that will be invoked only when

required.

One of another such major challenge is

Monitoring of such a highly distributed,

Even Driven Serverless system. Currently

AWS lambda can be monitored and

debugged only via AWS Cloud Watch

console logs. It is difficult to work with

Cloud Watch when data volumes is huge

and debugging becomes a difficult task

.However, some of the new emerging third

party tools like Dashbird , Epsagon and

IOPipe provide much advanced Serverless

FAAS Observability and Troubleshooting

capabilities and should be considered

while going Serverless .

Open Source Serverless FAAS platforms

offers an alternative approach to Vendor

provide Serverless FAAS platforms. These

platforms provide more flexibility for

Serverless implementations .These can be

deployed on Docker Swarm or Kubernetes.

Open Source Serverless offers a vendor

neutral approach, are flexible to cover

more scenarios without constraints and

offer function portability across cloud

platforms. This allows deployment on

premise, on laptop or any cloud of choice

or multi cloud environment. The projects

where the data size is more and execution

times are lengthy, Open Source fits better.

Enterprises already using Kubernetes(

either On premise or On public cloud

vendor managed service) to deploy their

existing micro services can easily migrate

to any of the Open source Serverless

platforms like Kubeless, Open FAAS,

KNative and Fission. This approach will

reuse the underlying Kubernetes service

with micro functions deployed over it.

Like any other technology, Serverless

architecture also comes with certain

constraints and pose challenges .There are

workarounds and workable solutions to

get over these limitations. The Serverless

functions provided by public Cloud

vendors come with a few constraints

like function execution time limitations,

Data payload size limitation and cold

starts. However, these limitations can be

easily overcome by various means and

combining Lambda with other services.

As an example AWS Lambda functions

currently have an execution time limitation


Conclusion

Serverless as one of the Key design goals

in any of their forthcoming transformation

programs and green field projects.

Serverless needs to be integrated as a core

part of an Enterprise Digital Strategy.

Serverless offers great advantages

for Startups. Stratups should think of

‘Serverless First Approach’ as their key

considerations in designing. This will help

them to launch large-scale innovative

ideas and robust features as per the current

market needs in no time, with reduced

workforce and reduced Cost.

A good Digital partner can help realize

the Serverless Dream of an Enterprise and

help them in their overall digital stack

modernization venture.

Proposed architecture of using Serverless

functions for handling and managing

business logic can play a very vital role in

implementing Digital Customer Experience

journeys. Core Enterprise journeys, Content

Management and e-Commerce scenarios

could easily gain advantage using our

Serverless Reference Architecture. This

architecture focuses on resolving the

major enterprise need today to provide

highly agile, lightweight and scalable

environment for always changing business

demands. On demand, granular scaling is

one of the key areas that it addresses which

is missing today in the PAAS platforms.

Enterprises could focus on providing the

actual functionalities and a world class

Customer experience leaving the worry of

Operational Administration to the Cloud

vendors providing the Serverless function

capability. There are some challenges

observed in Serverless approach.

However, with the developing advances

in Serverless technology, emerging Open

Source platforms and third party tools it is

possible to beat these minor hiccups and

go Serverless. The benefits of Serverless far

outweigh the small drawbacks that have

been observed.

The eventual future of Digital platforms is

‘Serverless First Architecture’ and as we see,

a large no. of the big players in the market

have already embraced and executed

this change to gain huge competitive

advantages.

Any big Enterprise should consider

modernizing their existing stack using

Serverless approach and incorporate


References

Credits

https://aws.amazon.com/lambda/?nc2=h_m1

https://docs.aws.amazon.com/lambda/latest/dg/lambda-services.html

https://aeqdigital.com/article/why-you-should-use-a-serverless-microservices-architecture/

https://www.bizety.com/2018/01/09/serverless-comparison-aws-lambda-vs-microsoft-azure-functions-vs-google-cloud-functions/

https://www.stratoscale.com/blog/cloud/aws-lambda-features-limitations-practical-examples/

https://techbeacon.com/app-dev-testing/event-driven-computing-best-practice-microservice-architecture

https://searchbusinessanalytics.techtarget.com/feature/How-big-data-analytics-tools-can-help-your-organization

https://codeburst.io/how-to-build-a-serverless-e-commerce-portal-48cdb6d49474

https://www.simform.com/ecommerce-app-using-serverless/

https://nickmchardy.com/2018/03/building-a-serverless-cms-using-aws-and-hugo.html

https://agileengine.com/serverless-architecture/

https://dzone.com/articles/serverless-retail

https://www.cosmicjs.com/

https://www.contentful.com/

1. MatHem case study : https://aws.amazon.com/solutions/case-studies/mathem/

2. ShutOut case study : https://serverless.com/blog/why-we-switched-from-docker-to-serverless/

3. CocaCola, BBC, AirBnb and Bustle.com Case Studies

https://northstack.com/why-companies-are-moving-to-serverless-cloud-technology/

https://aws.amazon.com/

https://dzone.com/

4. Example Use cases : https://aws.amazon.com/

5. Altassian case study : https://www.contentful.com/case-studies/atlassian/

6. Netflix case study : https://aws.amazon.com/solutions/case-studies/netflix-case-study/

https://dashbird.io/blog/serverless-case-study-netflix/

7. https://www.cosmicjs.com/

8. https://www.contentful.com/

9. https://www.webiny.com/

Archa ShuklaSenior Technology Architect, Infosys Digital

Archa Shukla has more than 15 years of experience in Delivering Complex Digital transformation programs across various industry domains Energy, Telecom, Manufacturing, Retail, Healthcare and Finance. She has led and delivered big programs leveraging public Cloud services and Serverless architecture that has integrations with real time IOT data.

About the Author


© 2020 Infosys Limited, Bengaluru, India. All Rights Reserved. Infosys believes the information in this document is accurate as of its publication date; such information is subject to change without notice. Infosys acknowledges the proprietary rights of other companies to the trademarks, product names and such other intellectual property rights mentioned in this document. Except as expressly permitted, neither this documentation nor any part of it may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, printing, photocopying, recording or otherwise, without the prior permission of Infosys Limited and/ or any named intellectual property rights holders under this document.

For more information, contact [email protected]

Infosys.com | NYSE: INFY Stay Connected

https://twitter.com/infosys

https://www.linkedin.com/company/infosys

https://www.youtube.com/user/Infosys

https://www.slideshare.net/infosys

https://www.infosys.com/

serverless in digital landscape an infosys point of view · source serverless platform based on...

Documents